The research plan details methods in statistical genetic that will aid in the study of the structure and function of biomolecular sequences. A molecular evolutionary, or comparative genomic, approach will be used, which allows for a correct treatment of correlations that arise from the common evolutionary origin of homologous DNA sequences. The following primary topics will be addressed: Multilocus Comparative Genomics. Statistical methods will be developed for determining whether or not nucleotide substitution rates at two or more loci are correlated. This type of approach will be useful in identifying mechanisms that affect mutation and substitution rates. Methods for partitioning total substitution rates into component sources (gene, species, etc) will be developed. The true evolutionary model for a gene is a reflection of its selective and functional constraints. By developing statistical methods for comparing models of sequence evolution between loci (or between regions within a single locus), functional studies of homologous sets of DNA sequences will be enhanced. Molecular Evolutionary Methods for Interacting Nucleotides. Phylogenetic comparisons of compensatory changes in RNA sequences have proven to be a powerful method for predicting RNA secondary structure. However, the methods that are usually employed lack any statistical basis. Statistical methods for analyzing these sequences will be developed. The methods will explicitly consider the correlated pattern of evolution among sites in secondary structure regions. A general approach for modeling "whole-sequence" dependencies will be explored. RNA secondary structure will receiver most of the initial attention, but the methodology developed can be modified for many types of dependencies.